OpenAI has started the training of its upcoming AI model to replace GPT-4 while setting up a safety and security committee to address critical issues. The move comes amidst resignations, with a focus on enhancing safety protocols and transparency.
OpenAI Forms Safety Committee as It Begins Training New AI Model
OpenAI announced on Tuesday, May 28, 2024, that it has initiated the training of its next frontier artificial intelligence model aimed to replace the GPT-4 system used in its ChatGPT chatbot. Concurrently, the San Francisco-based tech startup is establishing a new safety and security committee to advise its board on crucial safety and security issues concerning its operations and projects.
The formation of this committee comes amid heightened discussions about AI safety, following the resignations of prominent figures at OpenAI, including researcher Jan Leike and co-founder Ilya Sutskever. Leike criticized the company for prioritizing product development over safety.
The new committee comprises several key figures from OpenAI, including CEO Sam Altman and Chairman Bret Taylor. It also includes Quora CEO Adam D’Angelo and former Sony general counsel Nicole Seligman, alongside four technical and policy experts from within the company. Their initial task is to assess and enhance OpenAI’s safety protocols and issue recommendations to the board within 90 days. These recommendations will subsequently be made public in alignment with safety and security guidelines.
Despite the ongoing controversies, OpenAI maintains that its AI models lead the industry in both capability and safety. The company’s efforts come during a period of internal and external scrutiny, including lawsuits and concerns over AI’s potential societal impacts.
